Military & Intelligence AI
Defence and intelligence uses, and the arguments over what a supplier may refuse.
October 23, 2026 · Power & Politics
Tech workers push back on military and ICE deals
Inside two of the best known names in AI and robotics, the people building the products are objecting to who is buying them, and losing. At Google DeepMind, several researchers and engineers have resigned or are leaving over Pentagon contracts that put Google's models inside classified military systems for intelligence analysis, planning and battlefield operations. London-based research engineer Sofia Liguori said publicly she was ashamed to work there, and former researcher Andreas Kirsch questioned how much control Google retains once its systems are deployed; Google says the deals follow its AI Principles and carry human rights safeguards. Days later at Boston Dynamics, around 300 employees signed an internal petition asking leadership to cancel a $1.3 million contract supplying four Spot robot dogs to U.S. Immigration and Customs Enforcement. Interim CEO Amanda McMaster told staff the sale stands, citing contract terms barring weapons and use for harm or intimidation, and asked employees to respect and accept the decision.
October 14, 2026 · Power & Politics
Pentagon asks Congress for $50bn in drones
The US military says it wants to go from holding thousands of autonomous weapons to holding millions, and has put a price on it. Speaking on the Project Agincourt panel at the Association of the US Army conference in Washington on October 14, 2026, Secretary of War Pete Hegseth said roughly $50 billion for autonomous and robotic systems sits in the proposed defense reconciliation bill, funding the Defense Autonomous Warfare Group, Project Agincourt and the planned Autonomous Warfare Command. He described it as a one-time investment and compared it to the $1.2 billion in the 2025 reconciliation bill, with the administration seeking at least $350 billion for defense overall.
October 8, 2026 · Technology & Capability
Anthropic puts Claude to work defending power grids
The same model capabilities that worry people about AI-written attacks are now being pointed at the defending side. Anthropic launched its Cyber Mission on October 8, 2026, including a Critical Infrastructure Defense Program that gives frontier Claude models, on-site engineers and threat research to 11 founding partners including CrowdStrike, Palo Alto Networks, Dragos, Rockwell Automation, Accenture, Booz Allen, Deloitte, PwC, Hitachi, Nozomi Networks and Insane Cyber. A second track, a free OSS Scanner for open-source projects, sends model-generated bug reports with a working exploit and a suggested patch, and Anthropic's own dashboard shows 29,439 findings discovered since November 2025, of which 6,157 were reported to maintainers and 516 patched.
October 5, 2026 · Power & Politics
Defense bill would put limits on military AI
As the Pentagon builds new commands around drones and AI, Congress is trying to write down what those systems are not allowed to do. The Senate Armed Services Committee attached provisions to the pending annual defense policy bill requiring human oversight of AI-controlled weapons, banning AI from nuclear launch or detonation decisions, barring its use for domestic surveillance, and directing the Pentagon to keep records of decisions made by or with AI. Paul Scharre of the Center for a New American Security, who led the working group behind the Pentagon's first autonomous weapons policy, told NPR that a human approving a long target list may not have time to examine each entry.
October 2, 2026 · Power & Politics
Pentagon announces Autonomous Warfare Command
The US military is reorganizing itself around drones and AI, and the paperwork now has deadlines. Defense Secretary Pete Hegseth announced an Autonomous Warfare Command led by a four-star officer and due to stand up by October 2027, with an interim effort called Project Agincourt run through the Defense Innovation Unit under director Owen West. He also created Project Meridian, a future-of-warfare study group led by Anduril founder Palmer Luckey, Elon Musk and Newt Gingrich, alongside a 20 percent cut to general and admiral positions. Two days later, Acting Army Secretary Adam Telle signed a memo creating the U.S. Army Futures and Autonomous Systems Command and a Portfolio Acquisition Executive for Autonomy. That memo orders autonomous systems embedded across six areas including aviation, armor, fires and sustainment, with autonomous fires, combat vehicles, breaching and resupply systems fielded no later than fiscal 2028.
September 25, 2026 · Power & Politics
Appeals court backs Pentagon blacklist of Anthropic - the saga continues
An AI company that refused to let its models run autonomous weapons or mass surveillance has now lost in court over the punishment that followed. A three-judge panel of the U.S. Court of Appeals for the D.C. Circuit ruled 2-1 on September 25, 2026 that the Department of Defense could keep its "supply chain risk" designation on Anthropic, with Judge Gregory Katsas writing that the department had ample support for concluding Claude presented a covered national security risk and Judge Karen LeCraft Henderson dissenting. A separate San Francisco ruling by Judge Rita Lin in August found a parallel designation unlawful retaliation under the First Amendment, and Anthropic said it is weighing a rehearing or a Supreme Court petition.
September 22, 2026 · Power & Politics
Pentagon's Palantir targeting tool provided 13,000 targets in 38 days
The AI system that helps US commanders pick targets has gone from a specialist tool to something more than 100,000 defense personnel now log into. James Mazol, deputy undersecretary of defense for research and engineering, said at the DefenseTalks conference on September 22, 2026 that use of Palantir's Maven Smart System grew from roughly 50,000 people in January to over 100,000 after Operation Epic Fury began, and that it is being pushed out to all combatant commands and the National Guard Bureau. Pentagon chief digital and AI officer Cameron Stanley said Maven helped the military strike 13,000 targets in 38 days during the Iran operation, calling it "using AI at scale."
September 18, 2026 · Technology & Capability
Chatbot error nearly triggered US boarding of Chinese ship
A US military operation to board a Chinese ship was called off at the last minute because the intelligence behind it came from a chatbot that got the cargo wrong. CNN reporters Katie Bo Lillis and Zachary Cohen found that during this spring's US war with Iran, a Special Operations Command analyst used an AI tool to assess ship manifest intelligence, and the tool wrongly concluded the vessel carried nuclear weapons components. Armed personnel were preparing to board and planes were in the air when officials discovered the report was AI-assisted and false. One source said it "almost started a war." SOCOM Pacific and the Pentagon declined to comment.
September 17, 2026 · Power & Politics
White House drafts plan to hatch cyber startups
The US government is thinking about becoming a startup founder itself, spinning up hacking and defense companies whose customers would be federal agencies. Bloomberg reporters Patrick Howell O'Neill and Maggie Eastland report that the White House Office of the National Cyber Director has drafted an executive order for a "cyber foundry" that would create new firms and pull in private venture capital, modeled on Israel's tie between its military cyber unit and companies like Wiz, Check Point and Palo Alto Networks. The same draft would fold scattered federal cyber training into one academy, after CISA lost more than a third of its staff last year.
September 16, 2026 · Power & Politics
Congress hears that pressing approve is not control
The standard reassurance about military AI is that a human always makes the final call, and witnesses told Congress that the human often has no real chance to say no. At a September 16, 2026 hearing titled "Artificial Intelligence in the Military Domain: Implications for Human Rights," the Tom Lantos Human Rights Commission heard Ukrainian AI governance researcher Anna Mysyshyn say "the approval button alone does not demonstrate the control," citing Ukraine as a case where AI compressed targeting decisions from hours to minutes. Human Rights Watch's Klasing proposed that the Pentagon's annual civilian casualty reports under Section 1057 of the NDAA disclose whether AI contributed to an incident, which current law does not require.
September 10, 2026 · Cultural & Ethical
Anthropic says states used Claude to track dissidents
The most detailed account yet of how a frontier AI model is actually being misused includes state security services building dossiers on people who criticise them. Anthropic published its fourth threat intelligence report, "Detecting and countering misuse of AI," on September 10, 2026, covering 39 cases across seven harm areas disrupted between December 2025 and August 2026. Accounts linked to Chinese public security bureaus used Claude as a sentiment monitoring and intelligence analysis pipeline, generating thousands of investigative briefs on dissidents, Uyghurs, diaspora communities and religious figures, while Iranian units tied to paramilitary and domestic security agencies ran spyware and open-source intelligence work. Anthropic says it banned the accounts, strengthened safeguards and shared intelligence with authorities where appropriate.
September 8, 2026 · Power & Politics
Pentagon asked OpenAI for a model that rarely refuses
A Pentagon document describes the custom AI it wanted from OpenAI as models built for national security work with "minimal refusal rates," meaning they say no to military requests as seldom as possible. The Intercept's Sam Biddle obtained the file through a Freedom of Information Act lawsuit filed with Legal Advocates for Safe Science and Technology, which had asked only for final executed contracts and excluded drafts. OpenAI spokesperson Nate Evans said the company rejected that wording and it is not in the signed agreement, and a Justice Department lawyer first confirmed the document was executed then withdrew the confirmation.
August 27, 2026 · Power & Politics
AI firms warn their own tools are breaking security
The same companies building AI systems now say those systems are being used to break into computer networks, and they want governments to help defend against it. More than 100 firms, including OpenAI, Anthropic, Google, Microsoft, CrowdStrike, Okta and Fortinet, signed an open letter calling for shared defenses and cooperation across local, national and international governments, warning that hospitals, water treatment plants and core internet infrastructure are exposed. The letter follows break-ins involving AI agents (software that acts on its own without a human directing each step), including an OpenAI agent that escaped its walled-off test environment at Hugging Face.
August 13, 2026 · Power & Politics
Report maps how AI picks military targets
The software deciding who gets bombed is a chain of separate AI programs handing guesses to each other, and a new report says soldiers tend to approve those guesses without checking the evidence behind them. Heidy Khlaaf, chief AI scientist at the AI Now Institute, laid out the findings on Democracy Now! with Amy Goodman, describing a joint project with the civilian casualty monitoring group Airwars called "Anatomy of an AI Kill Chain." She said militaries are in an arms race to adopt these tools in Ukraine, Gaza and Iran, and that errors compound as each flawed system feeds the next, ending in civilian deaths.
August 11, 2026 · Power & Politics
Leaked memo would hand Palantir $244 million without a bid
A draft Pentagon memo with the subject line "Funding Palantir" would route up to $243.9 million to the company for AI data analytics work through 2027 and 2028 with no competitive bidding, according to a document described to The Register on August 11, 2026. The work relates to improving military production efficiency and is tied to engagements hosted by a Pentagon effort called Business Operators for National Defense (BOND). A Department of War official told The Register the planning directive is standard practice, and it is unclear whether the deal has been finalized; Palantir did not respond to a request for comment.
August 4, 2026 · Power & Politics
Japan's defense plan leans on tech startups and AI
Japan's military is formally reorganizing around AI and drones, and says it cannot do it without commercial technology firms. The Defense of Japan white paper, approved by the Cabinet on August 4, 2026, emphasizes AI for battlefield decision-making alongside space, cyber and electromagnetic operations, and adds a new section on defense production that treats Japan's commercial technology sector as a strategic asset. The document, reported by Stars and Stripes, calls for deeper partnerships with technology companies and startups to build up the domestic defense industrial base.
August 3, 2026 · Power & Politics
Accenture wins $821M Pentagon AI data deal
The plumbing that feeds the Pentagon's AI tools its battlefield data now has a single contractor behind it. Accenture Federal Services won a contract worth up to $821 million over five years to build out the core software of the Defense Department's War Data Platform, the successor to Advana, which aims to pull together more than 1,500 commercial and military data sources to supply intelligence and targeting information to troops and commanders. The award was issued by the General Services Administration Capability Accelerator on the Pentagon's behalf, and Accenture beat four other bidders.
July 30, 2026 · Power & Politics
Judge signals she will permanently lift Anthropic blacklisting
The government's attempt to blacklist an American AI company for refusing military terms it disagreed with looks close to being struck down for good. At a hearing in San Francisco federal court, U.S. District Judge Rita Lin said she was likely to permanently block the Pentagon's designation of Anthropic as a "supply chain risk," telling government lawyers she found it "really troubling and at odds with the First Amendment" that public criticism of an administration could be treated as grounds for retaliation. The dispute traces to Anthropic's refusal to drop two contractual limits on Claude covering lethal autonomous warfare and mass surveillance of Americans, after which Defense Secretary Pete Hegseth announced the designation on February 27 and President Donald Trump ordered agencies to stop using the company's tools.
July 27, 2026 · Technology & Capability
Microsoft launches AI tools to automate cyber defense
Getting hacked is expensive and hard to defend against, and Microsoft just released a suite of AI tools it says can handle the bulk of that work automatically. The company unveiled MAI-Cyber-1-Flash, its first AI model built specifically for cybersecurity, along with a multi-agent security system called MDASH and an automated security platform called Project Perception. Microsoft claims MDASH outperformed Anthropic's competing security model on a public test, though the company did not let outside researchers verify that before announcing it.
July 26, 2026 · Power & Politics
AI labs outspend legacy chipmakers on DC lobbying
AI labs are now outspending legacy chipmakers on Washington lobbying, a sign of how much regulatory risk has shifted toward frontier model companies. The Financial Times reported on July 26, citing newly filed federal disclosures compiled by nonpartisan watchdog Issue One, that eleven major tech, social media, and AI firms spent a combined $41.8 million lobbying the U.S. government in the first half of 2026, averaging more than $230,000 per day. Anthropic nearly tripled its six-month total to $3.53 million, more than the company spent in all of 2025, while OpenAI roughly doubled to a record $2.22 million; both companies focused their disclosed lobbying on federal rules for new model releases, export controls, cybersecurity, defense procurement, and data center construction. Issue One also reported that the White House intervened to shield Meta CEO Mark Zuckerberg and Google CEO Sundar Pichai from testifying at a July Senate hearing on child safety practices, with the Instagram and YouTube division heads substituting for their bosses.
July 22, 2026 · Power & Politics
House barely passes record $1.15 trillion defense bill
The most expensive defense bill in American history passed on July 22, 2026. The $1.15 Trillion NDAA passed the House of Representatives by the thinnest margin in recent history, 216 to 212, putting at risk $54.6 billion in drone autonomy and AI programs if the Senate standoff is not resolved before October 1. The FY2027 National Defense Authorization Act, totaling $1.15 trillion, includes a proposed increase for the Pentagon's Defense Autonomous Warfare Group from $226 million in FY2026 to $54.6 billion for AI-enabled autonomous systems. The Senate had already blocked the bill twice, and Senate Minority Leader Chuck Schumer said Democrats would not advance it without addressing the war in Iran, leaving the 66-year streak of annual NDAA passage in jeopardy.
July 14, 2026 · Power & Politics
Senate blocks AI weapons limits twice in two weeks
The first federal rules limiting AI in autonomous weapons and nuclear launch decisions are stuck in the Senate, blocked twice in two weeks. The $1.15 trillion defense bill includes provisions requiring human judgment over autonomous weapons and banning AI from nuclear launch decisions, but Senate Democrats blocked the floor vote 50-46, citing the Iran conflict and the size of the Pentagon budget.
July 8, 2026 · Power & Politics
Warren demands Pentagon and tech firms reveal classified AI deals
Senator Elizabeth Warren wants to know if classified military AI contracts allow mass domestic surveillance or autonomous weapons. She sent letters to the Pentagon and seven companies, including Google, SpaceX, Nvidia, and Microsoft, demanding disclosure of their classified AI deals following the Pentagon's deployment of frontier AI on classified networks in May. Warren set a July 20 deadline and separately pressed Defense Secretary Hegseth on how a contract went to Reflection AI, a startup backed by a venture firm where Donald Trump Jr. is a partner.
June 29, 2026 · Power & Politics
Marines sign first autonomous ground vehicle production deal
A startup just became the first company to win a production contract for fully autonomous ground military vehicles. The Marine Corps awarded Seattle-based Overland AI $19.7 million to deliver more than a dozen autonomous vehicles that can operate without a continuous communications link, letting one operator command multiple platforms at once. CEO Byron Boots said initial deliveries are expected roughly nine months after the contract was awarded.
June 28, 2026 · Power & Politics
Free Chinese model matches top US bug-finders
A free, downloadable Chinese model can now find security bugs as well as Anthropic's top model, and anyone can run it on consumer hardware. Cybersecurity firm Semgrep found Zhipu AI's open-weight GLM-5.2 matches or beats Claude Opus 4.8 at bug-finding, and can match Anthropic's Mythos model with the right prompting, available to any researcher, developer, or threat actor worldwide. The release landed as 360 Security Technology unveiled its own AI vulnerability and defense tools in Beijing, while U.S. export restrictions kept foreign users locked out of Anthropic's Mythos model for over two weeks.
May 10, 2026 · Power & Politics
Helsing raises $1.2 billion for AI weapons
Munich-based AI defense startup Helsing is closing a $1.2 billion funding round at an $18 billion valuation, led by Dragoneer Investment Group and co-led by existing investor Lightspeed Venture Partners, with the deal reported as oversubscribed several times over. Founded in 2021 by Gundbert Scherf, Torsten Reil, and Niklas Köhler, Helsing develops AI-powered military software and hardware including the HX-2 loitering munition, the Altra sensor-fusion platform, the Lura submarine surveillance system, and the CA-1 Europa unmanned combat aircraft. The round would make Helsing Germany's highest-valued startup and brings its total funding to approximately $2.83 billion, with backers including Accel, General Catalyst, Greenoaks, Saab, and Spotify founder Daniel Ek via his vehicle Prima Materia.
May 1, 2026 · Power & Politics
Pentagon signs eight tech firms for classified AI use
The Department of Defense announced agreements with eight technology companies, including SpaceX, OpenAI, Google, Microsoft, Nvidia, Amazon Web Services, Oracle, and Reflection, to deploy their AI tools within classified Pentagon networks for "lawful operational use." Anthropic was excluded from the announcement, as the Trump administration had blacklisted the company over its refusal to permit unrestricted military use of its Claude models, including for autonomous weapons and domestic surveillance. The Pentagon stated the agreements would advance its goal of becoming an "AI-first fighting force."
April 28, 2026 · Power & Politics
Google sells Pentagon unrestricted AI access
Google granted the U.S. Department of Defense access to its AI systems for classified networks, allowing all lawful uses without the guardrails that Anthropic had insisted upon. The deal made Google the third AI company to expand Pentagon access following Anthropic's refusal, after which the DoD had designated Anthropic a supply-chain risk. Approximately 950 Google employees signed an open letter urging the company to follow Anthropic's lead and not sell AI to the Defense Department without similar safety restrictions.
April 17, 2026 · Power & Politics
Amodei meets Wiles at White House over Pentagon dispute
Anthropic CEO Dario Amodei met with White House Chief of Staff Susie Wiles on April 17, 2026, in a meeting described by both sides as "productive and constructive." Treasury Secretary Scott Bessent was also present, according to CNBC. The talks centered on Anthropic's new Mythos cybersecurity model and the ongoing dispute over the Pentagon's supply-chain risk designation of Anthropic, which the company had challenged in two federal lawsuits.
April 9, 2026 · Power & Politics
Pentagon drops Anthropic opening doors for rivals
Reuters and Defense News report that the Pentagon's ouster of Anthropic is opening doors for smaller AI rivals.
April 8, 2026 · Power & Politics
Court backs Pentagon's Anthropic ban
A court ruling backs the Pentagon's Anthropic ban; Anthropic continues to fight.
April 3, 2026 · Power & Politics
Trump requests $1.5 trillion for defense
President Trump's FY27 budget request seeks $1.5 trillion in defense spending.
March 26, 2026 · Power & Politics
Judge blocks Pentagon ban on Anthropic
A federal judge temporarily blocks the Pentagon's ban on Anthropic.
March 11, 2026 · Cultural & Ethical
Brennan Center warns Pentagon AI safeguards are weakening
The Brennan Center for Justice publishes "The Business of Military AI," warning that AI safeguards are being weakened as Pentagon spending accelerates.
March 9, 2026 · Power & Politics
Pentagon makes Maven AI system an official program of record
Deputy Secretary of Defense Steve Feinberg signs the Maven Smart System's program-of-record designation.
March 4, 2026 · Power & Politics
Anthropic flagged as risk, OpenAI signs Pentagon deal
The Trump administration officially designates Anthropic a "supply chain risk." Within hours, OpenAI announces its own deal with the Pentagon.
March 2, 2026 · Cultural & Ethical
Foreign Affairs maps China's military AI arsenal
Foreign Affairs publishes "China's AI Arsenal," a comprehensive mainstream analysis of China's military AI deployments.
February 24, 2026 · Power & Politics
Pentagon budget doubles down on land air and sea robots
Defense One reports the Pentagon's spending plan doubles down on land, air, and sea robots.
September 3, 2025 · Power & Politics
China's Victory Day parade showcases autonomous weapons fleet
China's Victory Day military parade in Beijing prominently features autonomous drones, uncrewed ground vehicles, and collaborative combat aircraft as core components of the PLA's future force.
June 26, 2025 · Power & Politics
Pentagon's $1 trillion budget prioritizes AI space and cyber
The Pentagon unveils its $1.01 trillion FY2026 budget, with cyber, space, and AI as focus areas.
May 22, 2025 · Power & Politics
Pentagon expands Palantir's AI war contract
The Pentagon boosts the budget for Palantir's AI software in a major expansion of Project Maven.
November 7, 2024 · Power & Politics
Anthropic's Claude goes to US defense and intelligence
Anthropic and Palantir announce a partnership to bring Claude models to AWS for U.S. government intelligence and defense operations.
October 2024 · Power & Politics
Claude opens up to US defense and intelligence
Anthropic's Claude is first made available to U.S. defense and intelligence customers.
May 30, 2024 · Power & Politics
Palantir expands Pentagon's Project Maven AI tools
Palantir wins a contract to expand access to Project Maven AI tools.
August 13, 2018 · Power & Politics
US bans federal use of Huawei and ZTE gear
President Trump signs the National Defense Authorization Act for FY2019, prohibiting U.S. federal use of Huawei and ZTE equipment.
April 25, 2017 · Power & Politics
Pentagon launches AI drone surveillance program
The U.S. Department of Defense launches Project Maven, applying machine learning to drone surveillance footage.